Experimental studies in nonhuman primates and functional imaging studies in humans have underlined the critical role played by the prefrontal cortex (PFC) in working memory. However, the precise organization of the frontal lobes with respect to the different types of information operated upon is a point of controversy, and several models of functional organizations have been proposed. One model, developed by Goldman-Rakic and colleagues, postulates a modular organization of working memory based on the type of information processing (the domain specificity hypothesis). Evidence to date has focused on the encoding of the locations of visual objects by the dorsolateral PFC, whereas the ventrolateral PFC is suggested to be involved in processing the features and identity of objects. In this model, domain should refer to any sensory modality that registers information relevant to that domain--for example, there would be visual and auditory input to a spatial information processing region and a feature analysis system. In support of this model, recent studies have described pathways from the posterior and anterior auditory association cortex that target dorsolateral spatial-processing regions and ventrolateral object-processing regions, respectively. In addition, physiological recordings from the ventrolateral PFC indicate that some cells in this region are responsive to the features of complex sounds. Finally, recordings in adjacent ventrolateral prefrontal regions have shown that the features of somatosensory stimuli can be discriminated and encoded by ventrolateral prefrontal neurons. These discoveries argue that two domains, differing with respect to the type of information being processed, and not with respect to the sensory modality of the information, are specifically localized to discrete regions of the PFC and embody the domain specificity hypothesis, first proposed by Patricia Goldman-Rakic.